File size: 22,157 Bytes
6f3ebfa
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
"""
Training script for Vietnamese POS Tagger using CRF with Hydra config.

Supports 3 CRF trainers:
- python-crfsuite: Original Python bindings to CRFsuite
- crfsuite-rs: Rust bindings to CRFsuite (pip install crfsuite)
- underthesea-core: Underthesea's native Rust CRF implementation

Supports two datasets:
- VLSP 2013 POS (local, tab-separated word\tTAG format)
- UDD-1 (HuggingFace, Universal Dependencies)

Usage:
    python src/train_pos.py
    python src/train_pos.py data=udd1
    python src/train_pos.py model.trainer=crfsuite-rs
    python src/train_pos.py model.c1=0.5 model.c2=0.01
    python src/train_pos.py model.features.bigram=false

Feature ablation:
    python src/train_pos.py model.features.form=false
    python src/train_pos.py model.features.type=false
    python src/train_pos.py model.features.morphology=false
    python src/train_pos.py model.features.left=false
    python src/train_pos.py model.features.right=false
    python src/train_pos.py model.features.bigram=false
    python src/train_pos.py model.features.dictionary=false
"""

import logging
import platform
import re
import time
from abc import ABC, abstractmethod
from datetime import datetime
from pathlib import Path

import hydra
import psutil
import yaml
from omegaconf import DictConfig, OmegaConf
from sklearn.metrics import accuracy_score, classification_report

log = logging.getLogger(__name__)


# ============================================================================
# Feature Groups (for ablation study)
# ============================================================================

FEATURE_GROUPS = {
    "form":       ["T[0]", "T[0].lower"],
    "type":       ["T[0].istitle", "T[0].isupper", "T[0].isdigit", "T[0].isalpha"],
    "morphology": ["T[0].prefix2", "T[0].prefix3", "T[0].suffix2", "T[0].suffix3"],
    "left":       ["T[-1]", "T[-1].lower", "T[-1].istitle", "T[-1].isupper",
                   "T[-2]", "T[-2].lower"],
    "right":      ["T[1]", "T[1].lower", "T[1].istitle", "T[1].isupper",
                   "T[2]", "T[2].lower"],
    "bigram":     ["T[-1,0]", "T[0,1]"],
    "dictionary": ["T[0].is_in_dict", "T[-1,0].is_in_dict", "T[0,1].is_in_dict"],
}


def get_active_templates(features_cfg):
    """Build active feature template list from config."""
    templates = []
    for group_name, group_templates in FEATURE_GROUPS.items():
        if features_cfg.get(group_name, True):
            templates.extend(group_templates)
    return templates


def get_active_groups(features_cfg):
    """Return list of enabled group names."""
    return [g for g in FEATURE_GROUPS if features_cfg.get(g, True)]


# ============================================================================
# Utilities
# ============================================================================

def get_hardware_info():
    """Collect hardware and system information."""
    info = {
        "platform": platform.system(),
        "platform_release": platform.release(),
        "architecture": platform.machine(),
        "python_version": platform.python_version(),
        "cpu_physical_cores": psutil.cpu_count(logical=False),
        "cpu_logical_cores": psutil.cpu_count(logical=True),
        "ram_total_gb": round(psutil.virtual_memory().total / (1024**3), 2),
    }

    try:
        if platform.system() == "Linux":
            with open("/proc/cpuinfo", "r") as f:
                for line in f:
                    if "model name" in line:
                        info["cpu_model"] = line.split(":")[1].strip()
                        break
    except Exception:
        info["cpu_model"] = "Unknown"

    return info


def format_duration(seconds):
    """Format duration in human-readable format."""
    if seconds < 60:
        return f"{seconds:.2f}s"
    elif seconds < 3600:
        minutes = int(seconds // 60)
        secs = seconds % 60
        return f"{minutes}m {secs:.2f}s"
    else:
        hours = int(seconds // 3600)
        minutes = int((seconds % 3600) // 60)
        secs = seconds % 60
        return f"{hours}h {minutes}m {secs:.2f}s"


# ============================================================================
# Feature Extraction
# ============================================================================

def get_token_value(tokens, position, index):
    """Get token at position + index, with boundary handling."""
    actual_pos = position + index
    if actual_pos < 0:
        return "__BOS__"
    elif actual_pos >= len(tokens):
        return "__EOS__"
    return tokens[actual_pos]


def apply_attribute(value, attribute, dictionary=None):
    """Apply attribute transformation to a token value."""
    if value in ("__BOS__", "__EOS__"):
        return value
    if attribute is None:
        return value
    elif attribute == "lower":
        return value.lower()
    elif attribute == "istitle":
        return str(value.istitle())
    elif attribute == "isupper":
        return str(value.isupper())
    elif attribute == "isdigit":
        return str(value.isdigit())
    elif attribute == "isalpha":
        return str(value.isalpha())
    elif attribute == "is_in_dict":
        return str(value in dictionary) if dictionary else "False"
    elif attribute.startswith("prefix"):
        n = int(attribute[6:]) if len(attribute) > 6 else 2
        return value[:n] if len(value) >= n else value
    elif attribute.startswith("suffix"):
        n = int(attribute[6:]) if len(attribute) > 6 else 2
        return value[-n:] if len(value) >= n else value
    return value


def parse_template(template):
    """Parse a feature template like T[0].lower into indices and attribute."""
    match = re.match(r"T\[([^\]]+)\](?:\.(\w+))?", template)
    if not match:
        return None, None
    indices_str = match.group(1)
    attribute = match.group(2)
    indices = [int(i.strip()) for i in indices_str.split(",")]
    return indices, attribute


def extract_features(tokens, position, active_templates, dictionary=None):
    """Extract features for a token at given position."""
    features = {}
    for template in active_templates:
        indices, attribute = parse_template(template)
        if indices is None:
            continue
        if len(indices) == 1:
            value = get_token_value(tokens, position, indices[0])
            value = apply_attribute(value, attribute, dictionary)
            features[template] = value
        else:
            values = [get_token_value(tokens, position, idx) for idx in indices]
            if attribute == "is_in_dict":
                combined = " ".join(values)
                features[template] = str(combined in dictionary) if dictionary else "False"
            else:
                combined = "|".join(values)
                features[template] = combined
    return features


def sentence_to_features(tokens, active_templates, dictionary=None):
    """Convert token sequence to feature sequences."""
    return [
        [f"{k}={v}" for k, v in extract_features(tokens, i, active_templates, dictionary).items()]
        for i in range(len(tokens))
    ]


# ============================================================================
# Data Loading
# ============================================================================

def load_data(cfg):
    """Load dataset based on config."""
    if cfg.data.source == "local":
        return load_data_vlsp2013(cfg)
    elif cfg.data.source == "huggingface":
        return load_data_udd1(cfg)
    else:
        raise ValueError(f"Unknown data source: {cfg.data.source}")


def load_data_vlsp2013(cfg):
    """Load VLSP 2013 POS dataset (tab-separated word\\tTAG format)."""
    log.info("Loading VLSP 2013 POS dataset...")

    dataset_dir = Path(cfg.data.data_dir)

    def load_file(path):
        sentences = []
        current_tokens = []
        current_tags = []
        with open(path, encoding="utf-8") as f:
            for line in f:
                line = line.strip()
                if not line:
                    if current_tokens:
                        sentences.append((current_tokens, current_tags))
                        current_tokens = []
                        current_tags = []
                else:
                    parts = line.split("\t")
                    if len(parts) == 2:
                        current_tokens.append(parts[0])
                        current_tags.append(parts[1])
            if current_tokens:
                sentences.append((current_tokens, current_tags))
        return sentences

    train_data = load_file(dataset_dir / "train.txt")
    test_data = load_file(dataset_dir / "test.txt")

    train_tokens = sum(len(toks) for toks, _ in train_data)
    test_tokens = sum(len(toks) for toks, _ in test_data)

    log.info(f"Loaded {len(train_data)} train ({train_tokens} tokens), "
             f"{len(test_data)} test ({test_tokens} tokens) sentences")

    return train_data, None, test_data, {
        "dataset": "VLSP-2013-POS",
        "train_sentences": len(train_data),
        "train_tokens": train_tokens,
        "val_sentences": 0,
        "val_tokens": 0,
        "test_sentences": len(test_data),
        "test_tokens": test_tokens,
    }


def load_data_udd1(cfg):
    """Load UDD-1 dataset from HuggingFace."""
    from datasets import load_dataset

    log.info("Loading UDD-1 dataset...")
    dataset = load_dataset(cfg.data.dataset)

    def extract_sentences(split):
        sentences = []
        for item in split:
            tokens = item["tokens"]
            tags = item["upos"]
            if tokens and tags:
                sentences.append((tokens, tags))
        return sentences

    train_data = extract_sentences(dataset["train"])
    val_data = extract_sentences(dataset["validation"])
    test_data = extract_sentences(dataset["test"])

    train_tokens = sum(len(toks) for toks, _ in train_data)
    val_tokens = sum(len(toks) for toks, _ in val_data)
    test_tokens = sum(len(toks) for toks, _ in test_data)

    log.info(f"Loaded {len(train_data)} train ({train_tokens} tokens), "
             f"{len(val_data)} val ({val_tokens} tokens), "
             f"{len(test_data)} test ({test_tokens} tokens) sentences")

    return train_data, val_data, test_data, {
        "dataset": cfg.data.dataset,
        "train_sentences": len(train_data),
        "train_tokens": train_tokens,
        "val_sentences": len(val_data),
        "val_tokens": val_tokens,
        "test_sentences": len(test_data),
        "test_tokens": test_tokens,
    }


# ============================================================================
# Dictionary
# ============================================================================

def load_dictionary():
    """Load Viet74K + UTS Dictionary from underthesea package."""
    from underthesea.corpus.readers.dictionary_loader import DictionaryLoader
    from underthesea.datasets import get_dictionary
    dictionary = set()
    for word in DictionaryLoader("Viet74K.txt").words:
        w = word.lower().strip()
        dictionary.add(w)
    for word in get_dictionary():
        w = word.lower().strip()
        dictionary.add(w)
    return dictionary


# ============================================================================
# Trainer Abstraction
# ============================================================================

class CRFTrainerBase(ABC):
    """Abstract base class for CRF trainers."""

    name: str = "base"

    @abstractmethod
    def train(self, X_train, y_train, output_path, c1, c2, max_iterations, verbose=True):
        pass

    @abstractmethod
    def predict(self, model_path, X_test):
        pass


class PythonCRFSuiteTrainer(CRFTrainerBase):
    name = "python-crfsuite"

    def train(self, X_train, y_train, output_path, c1, c2, max_iterations, verbose=True):
        import pycrfsuite
        trainer = pycrfsuite.Trainer(verbose=verbose)
        for xseq, yseq in zip(X_train, y_train):
            trainer.append(xseq, yseq)
        trainer.set_params({
            "c1": c1, "c2": c2, "max_iterations": max_iterations,
            "feature.possible_transitions": True,
        })
        trainer.train(str(output_path))

    def predict(self, model_path, X_test):
        import pycrfsuite
        tagger = pycrfsuite.Tagger()
        tagger.open(str(model_path))
        return [tagger.tag(xseq) for xseq in X_test]


class CRFSuiteRsTrainer(CRFTrainerBase):
    name = "crfsuite-rs"

    def train(self, X_train, y_train, output_path, c1, c2, max_iterations, verbose=True):
        import crfsuite
        trainer = crfsuite.Trainer()
        trainer.set_params({
            "c1": c1, "c2": c2, "max_iterations": max_iterations,
            "feature.possible_transitions": True,
        })
        for xseq, yseq in zip(X_train, y_train):
            trainer.append(xseq, yseq)
        trainer.train(str(output_path))

    def predict(self, model_path, X_test):
        import crfsuite
        model = crfsuite.Model(str(model_path))
        return [model.tag(xseq) for xseq in X_test]


class UndertheseaCoreTrainer(CRFTrainerBase):
    name = "underthesea-core"

    def _check_trainer_import(self):
        try:
            from underthesea_core import CRFTrainer
            return CRFTrainer
        except ImportError:
            pass
        try:
            from underthesea_core.underthesea_core import CRFTrainer
            return CRFTrainer
        except ImportError:
            pass
        raise ImportError("CRFTrainer not available in underthesea_core.")

    def _check_tagger_import(self):
        try:
            from underthesea_core import CRFModel, CRFTagger
            return CRFModel, CRFTagger
        except ImportError:
            pass
        try:
            from underthesea_core.underthesea_core import CRFModel, CRFTagger
            return CRFModel, CRFTagger
        except ImportError:
            pass
        raise ImportError("CRFModel/CRFTagger not available in underthesea_core")

    def train(self, X_train, y_train, output_path, c1, c2, max_iterations, verbose=True):
        CRFTrainer = self._check_trainer_import()
        trainer = CRFTrainer(
            loss_function="lbfgs", l1_penalty=c1, l2_penalty=c2,
            max_iterations=max_iterations, verbose=1 if verbose else 0,
        )
        model = trainer.train(X_train, y_train)
        output_path_str = str(output_path)
        if output_path_str.endswith('.crfsuite'):
            output_path_str = output_path_str.replace('.crfsuite', '.crf')
        model.save(output_path_str)
        self._model_path = output_path_str

    def predict(self, model_path, X_test):
        CRFModel, CRFTagger = self._check_tagger_import()
        model_path_str = str(model_path)
        if hasattr(self, '_model_path'):
            model_path_str = self._model_path
        elif model_path_str.endswith('.crfsuite'):
            model_path_str = model_path_str.replace('.crfsuite', '.crf')
        model = CRFModel.load(model_path_str)
        tagger = CRFTagger.from_model(model)
        return [tagger.tag(xseq) for xseq in X_test]


def get_trainer(trainer_name: str) -> CRFTrainerBase:
    """Get trainer instance by name."""
    trainers = {
        "python-crfsuite": PythonCRFSuiteTrainer,
        "crfsuite-rs": CRFSuiteRsTrainer,
        "underthesea-core": UndertheseaCoreTrainer,
    }
    if trainer_name not in trainers:
        raise ValueError(f"Unknown trainer: {trainer_name}. Available: {list(trainers.keys())}")
    return trainers[trainer_name]()


# ============================================================================
# Metadata
# ============================================================================

def save_metadata(output_dir, cfg, data_stats, active_groups, active_templates, accuracy, hw_info, training_time):
    """Save model metadata to YAML file."""
    model_cfg = cfg.model
    metadata = {
        "model": {
            "name": "Vietnamese POS Tagger",
            "type": "CRF (Conditional Random Field)",
            "framework": model_cfg.trainer,
        },
        "training": {
            "dataset": data_stats.get("dataset", "unknown"),
            "train_sentences": data_stats["train_sentences"],
            "train_tokens": data_stats["train_tokens"],
            "val_sentences": data_stats["val_sentences"],
            "val_tokens": data_stats["val_tokens"],
            "test_sentences": data_stats["test_sentences"],
            "test_tokens": data_stats["test_tokens"],
            "hyperparameters": {
                "c1": model_cfg.c1,
                "c2": model_cfg.c2,
                "max_iterations": model_cfg.max_iterations,
            },
            "feature_groups": active_groups,
            "num_feature_templates": len(active_templates),
            "feature_templates": active_templates,
            "duration_seconds": round(training_time, 2),
        },
        "performance": {
            "test_accuracy": round(accuracy, 4),
        },
        "environment": {
            "platform": hw_info["platform"],
            "cpu_model": hw_info.get("cpu_model", "Unknown"),
            "python_version": hw_info["python_version"],
        },
        "created_at": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
        "author": "undertheseanlp",
    }

    metadata_path = output_dir / "metadata.yaml"
    with open(metadata_path, "w") as f:
        yaml.dump(metadata, f, default_flow_style=False, allow_unicode=True, sort_keys=False)
    log.info(f"Metadata saved to {metadata_path}")


# ============================================================================
# Main
# ============================================================================

@hydra.main(version_base=None, config_path="conf/pos", config_name="config")
def train(cfg: DictConfig):
    """Train Vietnamese POS Tagger using CRF."""
    total_start_time = time.time()
    start_datetime = datetime.now()

    log.info(f"Config:\n{OmegaConf.to_yaml(cfg)}")

    model_cfg = cfg.model

    # Resolve feature groups
    active_templates = get_active_templates(model_cfg.features)
    active_groups = get_active_groups(model_cfg.features)

    # Get trainer
    crf_trainer = get_trainer(model_cfg.trainer)

    # Determine output directory (relative to original cwd, not Hydra's output dir)
    original_cwd = Path(hydra.utils.get_original_cwd())
    output_dir = original_cwd / cfg.output
    output_dir.mkdir(parents=True, exist_ok=True)
    output_path = output_dir / "model.crfsuite"

    # Collect hardware info
    hw_info = get_hardware_info()

    log.info("=" * 60)
    log.info("POS Tagger Training")
    log.info("=" * 60)
    log.info(f"Dataset: {cfg.data.name}")
    log.info(f"Trainer: {model_cfg.trainer}")
    log.info(f"Features: {active_groups} ({len(active_templates)} templates)")
    log.info(f"Platform: {hw_info['platform']}")
    log.info(f"CPU: {hw_info.get('cpu_model', 'Unknown')}")
    log.info(f"Output: {output_dir}")
    log.info(f"Started: {start_datetime.strftime('%Y-%m-%d %H:%M:%S')}")
    log.info("=" * 60)

    # Load data
    train_data, val_data, test_data, data_stats = load_data(cfg)

    log.info(f"Train: {len(train_data)} sentences ({data_stats['train_tokens']} tokens)")
    if val_data:
        log.info(f"Validation: {len(val_data)} sentences ({data_stats['val_tokens']} tokens)")
    log.info(f"Test: {len(test_data)} sentences ({data_stats['test_tokens']} tokens)")

    # Build dictionary (if dictionary features enabled)
    dictionary = None
    if model_cfg.features.get("dictionary", True):
        log.info("Loading dictionary...")
        dictionary = load_dictionary()
        log.info(f"Dictionary: {len(dictionary)} words")

    # Prepare training data
    log.info("Extracting features...")
    feature_start = time.time()
    X_train = [sentence_to_features(tokens, active_templates, dictionary) for tokens, _ in train_data]
    y_train = [tags for _, tags in train_data]
    log.info(f"Feature extraction: {format_duration(time.time() - feature_start)}")

    # Train CRF
    log.info(f"Training CRF model with {model_cfg.trainer}...")
    crf_start = time.time()
    crf_trainer.train(
        X_train, y_train, output_path,
        model_cfg.c1, model_cfg.c2, model_cfg.max_iterations,
        verbose=True,
    )
    crf_time = time.time() - crf_start
    log.info(f"Model saved to {output_path}")
    log.info(f"CRF training: {format_duration(crf_time)}")

    # Evaluation
    log.info("Evaluating on test set...")

    X_test = [sentence_to_features(tokens, active_templates, dictionary) for tokens, _ in test_data]
    y_test = [tags for _, tags in test_data]

    y_pred = crf_trainer.predict(output_path, X_test)

    # Flatten for metrics
    y_test_flat = [tag for tags in y_test for tag in tags]
    y_pred_flat = [tag for tags in y_pred for tag in tags]

    accuracy = accuracy_score(y_test_flat, y_pred_flat)

    log.info(f"Accuracy: {accuracy:.4f}")
    log.info(f"Classification Report:\n{classification_report(y_test_flat, y_pred_flat)}")

    total_time = time.time() - total_start_time

    # Save metadata
    save_metadata(output_dir, cfg, data_stats, active_groups, active_templates, accuracy, hw_info, total_time)

    # Show examples
    log.info("=" * 60)
    log.info("Example predictions:")
    log.info("=" * 60)
    for i in range(min(3, len(test_data))):
        tokens = test_data[i][0]
        true_tags = y_test[i]
        pred_tags = y_pred[i]
        pairs_true = " ".join(f"{t}/{g}" for t, g in zip(tokens, true_tags))
        pairs_pred = " ".join(f"{t}/{g}" for t, g in zip(tokens, pred_tags))
        log.info(f"True: {pairs_true}")
        log.info(f"Pred: {pairs_pred}")

    log.info("=" * 60)
    log.info("Training Summary")
    log.info("=" * 60)
    log.info(f"Dataset: {cfg.data.name}")
    log.info(f"Trainer: {model_cfg.trainer}")
    log.info(f"Features: {active_groups} ({len(active_templates)} templates)")
    log.info(f"Model: {output_path}")
    log.info(f"Accuracy: {accuracy:.4f}")
    log.info(f"Total time: {format_duration(total_time)}")
    log.info("=" * 60)


if __name__ == "__main__":
    train()